1. Human muscle endurance is apparently enhanced during maximal voluntary contractions at short muscle lengths (McKenzie & Gandevia, 1987) but the ability of subjects to activate muscles fully at short lengths has not been established. Therefore this study examined the voluntary capacity to activate muscles fully at control (near resting) lengths and at decreased muscle lengths. Changes in mechanical properties of twitch responses to electrical stimulation of relaxed muscles at short muscle lengths were also documented. The abductor digiti minimi, elbow flexors and tibialis anterior were studied in five subjects. 2. For the three muscle groups, the mean reduction in twitch force between the control and short muscle lengths ranged from 46-51%. AT the short length there was a 9-13% reduction in the contraction time and a 21-27% reduction in the half-relaxation time. Maximal voluntary force declined by 21-49% at the short muscle length. A reduction in muscle length produced a shift to the right of the force-frequency curve as determined by brief trains of electrical stimuli. 3. During maximal efforts single or brief trains of two to four supramaximal stimuli, delivered to the parent nerve or motor point, failed to increase the force at a latency appropriate for onset of a muscle twitch in some but not all attempts. Each subject achieved 'maximal activation' of the muscle in a similar proportion of attempts at the control and short muscle lengths. 4. These results suggest that maximal voluntary activation of motoneurone pools is possible at short muscle lengths and that the central nervous system is able to maintain the discharge of motoneurones close to 'fusion' frequency despite a decrease in the temporal characteristics of the isometric twitch.